In lieu of an abstract, here is a brief excerpt of the content:

A P P E N D I X A Assessing the Validity of the CMAG Tracking Data T O ASSESS THE VALIDITY of the 2000 CMAG data, we visited eight television stations in 2001 in five different markets: WJRT in Flint, Michigan; WISN in Milwaukee, Wisconsin ; WTAE and KDKA in Pittsburgh, Pennsylvania; WRIC in Richmond, Virginia; and WJLA, WRC, and WUSA in Washington, D.C. At each station, we obtained photocopies of a sample of billing invoices prepared for various political advertising buyers. Each invoice included the name of the advertiser (e.g., Gore/Lieberman, Inc.; the Virginia Democratic Party) and a list of the dates on and times at which spots aired. Next, we matched the spots listed on the invoices with those listed in the CMAG dataset. With one exception, we found the tracking data to be highly accurate (see Table A.1). At WJRT in Flint, for example, we collected a total of forty-one invoices representing 618 individual ad airings. Six spots listed on the invoices were not found in the CMAG dataset; the remaining 612 were listed accurately to within one minute. As for the six discrepancies, it is important to note that these spots may, in fact, never have been aired, in which case the error would be one of inadequate record-keeping on the part of the station, not an error in the CMAG data. At WISN in Milwaukee, the error rate was slightly higher. On the thirty-nine TABLE A.1 Assessments of Tracking Data Accuracy Percentage Percentage correct within off by more Percentage Television station 1 minute than 1 minute not located Total spots WJRT Flint 99 0 1 618 WISN Milwaukee 97 <1 2.3 977 KDKA Pittsburgh 80 0 20 308 WTAE Pittsburgh 97 0 3 773 WRIC Richmond 96 2 2 161 WJLA Washington, D.C. 98 1 1 698 WRC Washington, D.C. 99 <1 0 363 WUSA Washington, D.C. 97 0 3 1433 146 A P P E N D I X A invoices collected, 977 airings were documented. For five of those, the time stamp given by CMAG differed from the television station records by more than one minute, and we were unable to locate twenty-two of the airings recorded on the invoices. Still, more than 97 percent of the time, the CMAG data and the station invoices agreed within one minute as to when a spot aired. At WTAE, WRIC, and the three Washington, D.C., stations, accuracy was similarly high. The one trouble spot was KDKA in Pittsburgh. For reasons unknown to us, we were unable to locate in the CMAG tracking data 62 of the 308 spots listed on the invoices obtained from the television station. The missing spots, however, were not related to any candidate or any period of time in particular, suggesting that in even this one outlier case, the error was essentially random. In all, we were able to locate 96 percent of the airings listed in the station invoices, and more than 95 percent of cases were accurate to within one minute. In sum, there seems to be little need for researchers to invest the time and resources into traveling to television stations to collect advertising data; advertising tracking data provide an accurate and reliable substitute. ...

Share